R2T2 - Research References Tracking Tool

The Research References Tracking Tool (R2T2) aims to fill the last remaining gap into the circle of open research, enabling not just publications to cite data or software (still a work in progress), but also the latter to cite the research articles containing the theory it codes, or the datasets it uses.

Some of the benefits of using R2T2 in your project are:

• Facilitate giving credit to all the works software is based on.
• Promote those works’ visibility and impact.
• Promote the transparency of the code: bi-directional link between theory and the specific code that implements it.
• Facilitate code maintenance and improve its sustainability.

Installation

R2T2 is available in PyPI, so to install it just run:

pip install R2T2

How it works

R2T2 works by decorating those functions, classes or methods where particular algorithms described in a paper are implemented or data stored in a repository is used. General execution of code silently passes these decorators, but remembers how and where they were called. The decorators include a short description of the thing being reference, and the reference itself in any sensible format.

from r2t2 import add_reference
...
@add_reference(short_purpose="Original implementation of R2T2", 
                  reference="Diego Alonso-Álvarez, et al."
                            "(2018, February 27). Solcore (Version 5.1.0). Zenodo."
                            "http://doi.org/10.5281/zenodo.1185316")
def my_great_function():
    pass

Several references can be added by stacking multiple @add_reference decorators.

@add_reference(short_purpose="some comment",  reference="Reference 1")
@add_reference(short_purpose="another comment",  reference="Reference 2")
def my_great_function():
    pass

There are two methods of using this information:

Runtime tracker

Which markers were passed when running a particular script my_script.py can be recalled with:

$ python -m r2t2 my_script.py

This prints a list of markers passed in the script run and recursively in any dependency used by the program. Input arguments needed by the script can be added after its name.

$ python -m r2t2 my_script.py arg1 arg2

Static tracker

Alternatively, R2T2 can be used to provide a list of all references that a given package is based on (i.e. ALL the add_reference decorators it contains ) and not just those crossed by a particular run of a script using the package.

For using this method, simply run in the terminal:

$ python -m r2t2

which will scan all the python files recursively starting in the current directory. By default, it prints the results in the terminal. To analyse a single file, use the flag -s (from "static") to prevent r2t2 to treat it as a script to run:

$ python -m r2t2 -s my_script.py

To have more control on what is scanned, the format of the output and where the output is written:

$ python -m r2t2 -f markdown -o docs/list_of_references.md some/subdirectory 

The contents of the output will be organised by decorated object in the order they were encountered and contain the line where the decorator was found, a link to that location, and the list of the short purposes and the references itself:

Referenced in: roasted_chicken  
Source: [tests/test_r2t2.py](tests/test_r2t2.py:7)  
Line: 7

	[1] Roasted chicken recipe - Great British Roasts, 2019  

Prior art

R2T2 is based in part on work done by Markus Führer at Imperial College London, as part of the Solcore project.